Information optimization method, device, server and storage medium

Abstract

The embodiment of the present invention discloses an information optimization method, device, server and storage medium, wherein the method comprises: obtaining request information sent by a client, and determining response information corresponding to the request information from a preset information database; sending the response information to the client, and obtaining feedback information input based on the response information from the client; determining satisfaction with the response information according to the feedback information, and optimizing the response information when the satisfaction is less than a preset satisfaction threshold to obtain optimized response information, wherein the satisfaction determined according to the optimized response information is greater than or equal to the satisfaction threshold, which can improve the updating efficiency of information data in the server.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an information optimization method, device, server and storage medium. Background Art

[0002] Currently, when the original data of the server is updated, it is generally updated based on crowdsourcing, user surveys or manually collected information data. Collecting information data through crowdsourcing or user surveys consumes a lot of manpower and material resources, resulting in excessively high costs for updating the original data. Moreover, since the efficiency of collecting information data through crowdsourcing or user surveys is low, the efficiency of updating the original data is low. Therefore, how to quickly collect information data and effectively update the original data based on the collected information data has become a current research hotspot. Summary of the invention

[0003] The embodiments of the present invention provide an information optimization method, device, server and storage medium, which can improve the updating efficiency of information data in the server.

[0004] On the one hand, an embodiment of the present invention provides an information optimization method, the method comprising:

[0005] Obtaining the request information sent by the client, and determining the response information corresponding to the request information from a preset information database;

[0006] Sending the response information to the client, and obtaining feedback information input based on the response information from the client;

[0007] The satisfaction level for the response information is determined based on the feedback information, and when the satisfaction level is less than a preset satisfaction level threshold, the response information is optimized to obtain optimized response information, wherein the satisfaction level determined based on the optimized response information is greater than or equal to the satisfaction level threshold.

[0008] On the other hand, an embodiment of the present invention provides an information optimization device, the device comprising:

[0009] An acquisition unit, used for acquiring request information sent by a client;

[0010] A determination unit, configured to determine response information corresponding to the request information from a preset information database;

[0011] A sending unit, configured to send the response information to the client;

[0012] The acquisition unit is further configured to acquire feedback information input based on the response information from the client;

[0013] An optimization unit is used to determine the satisfaction level of the response information based on the feedback information, and when the satisfaction level is less than a preset satisfaction level threshold, optimize the response information to obtain optimized response information, wherein the satisfaction level determined based on the optimized response information is greater than or equal to the satisfaction level threshold.

[0014] In another aspect, an embodiment of the present invention provides a server, including a processor, a memory, and a communication interface, wherein the processor, the memory, and the communication interface are connected to each other, wherein the memory is used to store computer program instructions, and the processor is configured to execute the program instructions, performing the following steps:

[0015] Obtaining the request information sent by the client, and determining the response information corresponding to the request information from a preset information database;

[0016] Sending the response information to the client, and obtaining feedback information input based on the response information from the client;

[0017] The satisfaction level for the response information is determined based on the feedback information, and when the satisfaction level is less than a preset satisfaction level threshold, the response information is optimized to obtain optimized response information, wherein the satisfaction level determined based on the optimized response information is greater than or equal to the satisfaction level threshold.

[0018] On the other hand, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes the method of the first aspect.

[0019] In an embodiment of the present invention, a server can obtain request information sent by a client, thereby determining response information corresponding to the request information from a preset information database. After the server sends the determined response information to the client, feedback information input based on the response information can be obtained from the client. After the server determines satisfaction with the response information based on the feedback information, when it is determined that the satisfaction is less than a preset satisfaction threshold, the response information can be optimized to obtain response information that can improve user satisfaction. This realizes automatic updating of response information in the server based on user feedback, and can improve the updating efficiency of the response information. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0021] Figure 1a is a schematic block diagram of an information optimization system provided by an embodiment of the present invention;

[0022] Figure 1b This is a system architecture diagram of updating response information based on a knowledge graph information database provided by an embodiment of the present invention;

[0023] Figure 1c It is a schematic diagram of a flow chart of updating response information by an information optimization system based on a knowledge graph provided by an embodiment of the present invention;

[0024] Figure 2 It is a flowchart of an information optimization method provided by an embodiment of the present invention;

[0025] Figure 3 is a flow chart of an information optimization method provided by another embodiment of the present invention;

[0026] Figure 4 It is a schematic diagram of a process of generating guidance information provided by an embodiment of the present invention;

[0027] Figure 5 is a schematic diagram of a flow chart of generating guidance information provided by another embodiment of the present invention;

[0028] Figure 6 This is an interactive interface diagram for information interaction with a server provided by an embodiment of the present invention;

[0029] Figure 7 is a schematic block diagram of an information optimization device provided by an embodiment of the present invention;

[0030] Figure 8 It is a schematic block diagram of a server provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to optimize the original response information in the server during the interaction between the server and the client, so that the optimized response information is more in line with the dialogue scenario between the server and the client, and at the same time improve the accuracy of the response information, an information optimization method is proposed in an embodiment of the present invention, which can optimize the response information. The information optimization method can be applied to Figure 1aIn the information optimization system shown, the information optimization system includes a server 10 and a client 11. The server 10 is preset with an information database. The preset information database can respond to the request information obtained from the client 11. The server 10 can send the response information determined by the information database to the client 11, thereby realizing the information interaction between the server 10 and the client 11. Based on the information interaction between the server 10 and the client 11, the server 10 can realize the collection of information data and the update of response information. In one embodiment, the server 10 and the client 11 can first establish a communication connection through a preset communication interface. When the server 10 interacts with the client 11, it can first generate a system account for information exchange, so that information interaction can be performed based on the instant communication account (or user account) corresponding to the system account and the client 11. That is, the server 10 can obtain the request information by starting the instant communication account of the client 11. In one embodiment, the request information obtained by the server 10 from the client 11 can be, for example, question information, poetry information, or couplet information.

[0032] In one embodiment, after obtaining the request information, the server 10 may determine the response information corresponding to the request information based on the information database, so that the obtained response information may be sent to the client 11. After receiving the response information, the client 11 may send the feedback information input by the user based on the response information to the server 10. The server 10 determines whether to optimize the response information based on the user's satisfaction with the response information indicated by the feedback information, wherein when the server 10 determines that the user's satisfaction with the feedback information is less than a preset satisfaction threshold, it determines to optimize the response information, thereby obtaining the optimized response information. After the server 10 determines the response information corresponding to the request information, and after determining to optimize the response information to obtain the optimized response information, the server 10 may collect information data, wherein the collected information data includes one or more of the request information input by the client 11, the response information corresponding to the request information, and the optimized response information obtained by optimizing the response information.

[0033] In one embodiment, the information database preset in the server 10 includes a knowledge graph, wherein the knowledge graph is a data storage structure, and the knowledge graph includes multiple storage nodes and multiple edges, any of the storage nodes is used to store a keyword, and any two storage nodes with an associated relationship are connected by an edge, and the edge connecting the any two storage nodes is used to describe the association relationship between the keywords stored in the storage nodes. In one embodiment, see Figure 1bThe system architecture diagram for updating response information based on the information database of the knowledge graph is shown. As shown in the figure, the system architecture takes the information pair, knowledge graph, chat module and chat robot engine (chatterbot) based on machine learning as the main body of the information database, and encapsulates the information pair, knowledge graph, chat module and chatterbot in the preset communication interface, so that after the server 10 obtains the request information of the client 11, it can find the corresponding response information from the information database based on the request information, wherein the information pair stores multiple preset request information and the preset response information for any of the preset request information. When the server 10 cannot determine the response information corresponding to the request information based on the knowledge graph, it can find the preset response information for the request information from the information pair to ensure that the server can output the response information corresponding to the request information to the client.

[0034] When the server 10 and the client 11 communicate through a preset communication interface, the server 10 can have a single chat with the generated system account and the instant messaging account running on the client 11, or the server 10 can also pull the system account generated by the server 10 into a chat group through the instant messaging account running on the client 11 to have a group chat, thereby realizing information interaction between the server 10 and the client 11. The server 10 can guide the user to inquire and dialogue about relevant field information, and after receiving the request information input by the user, based on the information pairs in the information database, the knowledge graph, the chat module and the chatterbot, it can determine the response information for the request information, and output the response information to the client 11 to form a multi-round dialogue process. During the conversation between the server 10 and the user, the server 10 can obtain a large amount of information data, so that the obtained information data can be processed through natural language processing technology (NLP), and the processed information data can be fed back to the information database, so that the information database can perform machine learning based on the obtained information data, thereby realizing the update of the information database, and the response information output by the updated information database based on the request information obtained from the client is more in line with the user's needs. Among them, the preset interface service can be, for example, a web server hypertext transfer protocol (http) interface service, and the information data collected by the server 10 includes all information data when interacting with the client 11.

[0035] In one embodiment, see Figure 1c , is a flow chart of updating response information by an information optimization system based on a knowledge graph proposed in an embodiment of the present invention, such as Figure 1cAs shown, the above-mentioned information optimization system can first select a professional field of the knowledge graph or a field that needs to update the knowledge graph as the target field. For example, the target field of the knowledge graph can be selected as the World Cup field, or the male singer field or the female singer field, etc., wherein the keywords in the knowledge graph belonging to the same field are all related to the subject indicated by the field. After determining the target field from the knowledge graph, guidance information can be generated based on the target field. The guidance information is used to guide the user to send request information related to the target subject indicated by the target field to the server through the client. In one embodiment, the server can generate the guidance information through a preset guidance information template or a related artificial intelligence (AI) model, so that the instant messaging service interface (IM service interface) can be used to access the client for interaction, so as to obtain the request information sent by the user through the client.

[0036] After the server receives the request information sent by the user through the client, it can further extract information text from the request information, and the information text is the request keyword in the request information. For example, if the request information received by the server is "Where is the first World Cup held?", the request keywords extracted based on the request information include: first, World Cup, and venue. After the server extracts the request keywords from the received request information, it can call the core module in the question-and-answer system to return the response information corresponding to the request information (i.e., respond to the request information). Among them, the core module includes information pairs, a chat module, an artificial template, and a natural language understanding (NLU) module. When calling the core module to return the response information corresponding to the user's request information, the response keyword corresponding to the request keyword can be first searched from the knowledge graph, such as searching for the response keywords corresponding to the first, World Cup, and venue mentioned above, so that the response information corresponding to the request information can be generated based on the found response keywords and response templates. In addition, if the response keyword corresponding to the request keyword cannot be found in the knowledge graph, similar information matching the request information can be found from the multiple preset request information stored in the information pair, and the response information preset for the similar information can be determined from the information pair, so that the response information preset for the similar information can be output as the response information corresponding to the request information input by the client; or, if similar information matching the request information cannot be found from the information pair, the server can generate the response information corresponding to the request information through an artificial template to realize chatting with the user on the client side, or the server can also start the chatting mode with the user through NLU technology and output the response information corresponding to the request information.

[0037] In one embodiment, after the server determines the response information corresponding to the request information based on the request information obtained from the client, the data collection module may record the request information and the response information corresponding to the request information, and send the response information to the client, thereby obtaining the feedback information input by the client based on the response information. If the feedback information is successfully obtained, an information tag is added to the response information, and the information tag includes a user tag and an expert tag. The feedback information is used to indicate the satisfaction of the user corresponding to the client with the response information. Based on the received feedback information, the server may determine whether to add a user tag to the response information, and the user tag indicates the user's satisfaction with the response information. Further, the server may collect feedback information input by multiple users based on the response information of the same request information, and after sorting the feedback information of the multiple users for the response information, add an expert tag to the response information, and the expert tag is used to indicate the satisfaction of multiple users in the user set with the response information.

[0038] In one embodiment, if the server does not obtain feedback information based on the response information input from the client, then no information tag is added to the response information. When it is determined that the response information does not have an information tag, it can be determined whether the response information has been added with an information tag based on the user's historical feedback information. If it is determined that the information tag has been added, the existing information tag is obtained. Alternatively, if it is determined that the response information does not have an information tag, a clustering operation is performed on the request information to determine the information set to which the request information belongs, and the target information tag in the information set that has the same information category as the request information is used as the information tag of the request information.

[0039] After the server determines the expert annotation corresponding to the response information, or determines the information set to which the request information belongs, the request information, or the response information, or the request information and the response information can be divided into training data related to the knowledge graph, or training data related to the core module using natural language processing (NLP) technology, thereby updating the knowledge graph and the core module. Specifically, the training data related to the knowledge graph obtained based on the classification can update the entities indicated by the knowledge graph, the relationships and attributes between the entities, or add the entities that were not originally stored in the knowledge graph, the relationships and attributes between the entity and other entities to the knowledge graph; the server can perform incremental learning based on the obtained training data related to the core module, enrich the data of the manual module and the chat module, and make the interaction between the server and the client smoother.

[0040] See also Figure 2 , is a schematic flow chart of an information optimization method proposed in an embodiment of the present invention, the information optimization method can be applied to the above-mentioned server configured with an information database, such as Figure 2 As shown, the method may include:

[0041] S201, obtaining request information sent by a client, and determining response information corresponding to the request information from a preset information database.

[0042] In one embodiment, the server and the client can first establish a communication connection based on a preset communication interface, and the preset communication interface can be, for example, an http interface. After the client and the server establish a communication connection, the server can automatically generate a system account for accessing the client. When the user has a need to interact with the server, the user can start the client through the user account. After starting the client, the user can click on the system account corresponding to the server in the client to open an interactive interface for information interaction with the server. Based on the interactive interface, the user can enter request information so that the server can obtain the request information entered by the user through the client to achieve information interaction between the client and the server. In one embodiment, the server can obtain the request information sent by the client in real time, and can also obtain the historical unprocessed request information sent by the client, wherein the request information can be question information, poetry information, couplet information, etc. The question information refers to information that requires an answer or explanation. The information sentence pattern of the question information can be a question sentence pattern, such as "Where is the venue for the first World Cup?", etc., or a declarative sentence pattern, such as "Today is a good weather", etc., or an exclamatory sentence pattern, etc.

[0043] In one embodiment, after the server obtains the request information sent by the client, if the request information is in text form, the server directly determines the response information corresponding to the request information from a preset information database based on the request information; if the server obtains the request information in voice or video form from the client, the server can first determine the text information form corresponding to the request information in voice or video form, and then determine the response information corresponding to the text information form from the preset information database.

[0044] In one embodiment, when the server determines the response information corresponding to the request information, if the response keywords of the request information can be determined from the knowledge graph based on the request information, the response information corresponding to the request information is generated based on the response keywords and the response template in the information database. For example, if the request information is "Where was the first World Cup held?", the response keywords of the request information can be found from the knowledge graph related to the World Cup. The response keywords can be, for example, 1930 and Uruguay. Substituting the response keywords into the response template, the generated response information can be, for example, "The first World Cup was held in Uruguay in 1930."

[0045] In one embodiment, if the server cannot find the response keyword corresponding to the request information from the knowledge graph, the server will find the request information from the information pair of the information database, or similar information whose matching degree with the request information meets the preset matching degree threshold, and use the found request information or the response information corresponding to the similar information as the response information corresponding to the request information sent by the client. For example, if the keywords stored in each node in the knowledge graph are related to the World Cup, and the request information received from the client is "What was the date of New Year's Eve last year?" Since the response keyword corresponding to the request information cannot be found in the knowledge graph related to the World Cup, the request information or similar information matching the request information can be found in the information pair. If the similar information matching the request information found in the information pair is "When was New Year's Eve in 2018?", the response information corresponding to the similar information can be output to the client.

[0046] In one embodiment, if the server can neither determine the response keywords corresponding to the request information from the knowledge graph nor find similar information similar to the request information, it can start a chat mode with the user through a chat module or an NLU module to enhance the interactive experience between the user and the server. For example, when the request information is "Do you know the difference between fatherly love and motherly love?", it is impossible to determine the response keywords corresponding to the request information from the knowledge graph related to the World Cup, nor is it possible to find similar information similar to the request information from the information pair. Then, the response information sent to the client based on the determination of the chat module or the NLU module may be, for example, "Fatherly love and motherly love are both love that is given to us with all their heart."

[0047] In one embodiment, after the server determines the response information corresponding to the request information sent by the client, it may turn to execute step S202.

[0048] S202: Send the response information to the client, and obtain feedback information input based on the response information from the client.

[0049] In one embodiment, after determining the response information corresponding to the request information, the server may output the response information to the above-mentioned interactive interface for information interaction with the client, so that the response information can be displayed to the user through the interactive interface. After the client displays the received response information to the interactive interface, the user can judge the correctness of the response information to indicate the user's satisfaction with the response information, wherein the user can input feedback information through the client after judging the correctness of the response information to indicate the user's satisfaction with the response information. For example, if the request information sent by the client to the server is "Where was the first World Cup held?", the response information sent by the server to the client based on the request information is "The first World Cup was held in Uruguay in 1930", and the user determines that he is satisfied with the response information according to the judgment, then if the feedback information input through the client is "correct", it indicates that the user is satisfied with the feedback information, and the corresponding satisfaction is 100%.

[0050] In one embodiment, after determining the feedback information corresponding to the response information, the server may proceed to execute step S203.

[0051] S203, determining the satisfaction level for the response information according to the feedback information, and optimizing the response information when the satisfaction level is less than a preset satisfaction level threshold to obtain optimized response information.

[0052] In one embodiment, the server may determine the user's satisfaction with the response information according to the feedback information received from the client for the response information. Specifically, after receiving the feedback information, the server may analyze the feedback information to determine the user's emotional orientation towards the response information expressed by the feedback information. If the feedback information includes keywords such as correct, yes, right, good, etc., the user's emotional orientation towards the response information is determined to be a positive emotional orientation. If the feedback information includes keywords such as no, wrong, wrong, etc., the user's emotional orientation towards the response information is determined to be a negative emotional orientation. Among them, the positive emotional orientation indicates that the user is highly satisfied with the response information, and the negative emotional orientation indicates that the user is less satisfied with the response information. The server may preset a satisfaction threshold to distinguish between the satisfaction corresponding to the positive emotional orientation and the satisfaction corresponding to the negative emotional orientation, so that after the server determines the satisfaction corresponding to the response information according to the feedback information, the server may compare the satisfaction with the preset satisfaction threshold, thereby determining whether the satisfaction with the response information is less than the preset satisfaction threshold. If it is determined to be less than, it indicates that the user is not satisfied with the response information currently output by the server, thereby determining that the response information needs to be optimized to obtain the optimized response information. The satisfaction level determined according to the optimized response information is greater than or equal to a satisfaction level threshold preset by the server, and the preset satisfaction level threshold may be, for example, 85% or 90%.

[0053] When the service optimizes the response information, the response information to be optimized and the request information corresponding to the response information to be optimized can be sent to the developer, who can optimize the response information to be optimized according to the request information and input the optimized response information into the service. Alternatively, the server can also obtain the response information corresponding to the request information from the Internet based on the Internet, thereby optimizing the response information to be optimized, and input the optimized response information into the server, so that when the server obtains the request information from the client again, it can output the optimized response information to the client, thereby improving the user's satisfaction with the response information, improving the accuracy and practicality of human-computer interaction, and improving the user experience.

[0054] In an embodiment of the present invention, a server can obtain request information sent by a client, thereby determining response information corresponding to the request information from a preset information database. After the server sends the determined response information to the client, feedback information input based on the response information can be obtained from the client. After the server determines satisfaction with the response information based on the feedback information, when it is determined that the satisfaction is less than a preset satisfaction threshold, the response information can be optimized to obtain response information that can improve user satisfaction. This realizes automatic updating of response information in the server based on user feedback, and can improve the updating efficiency of the response information.

[0055] See also Figure 3 , is a schematic flow chart of an information optimization method proposed in another embodiment of the present invention, and the information optimization method can also be applied to the above-mentioned server configured with the information database, such as Figure 3 As shown, the method may include:

[0056] S301, obtaining request information sent by a client, and determining response information corresponding to the request information from a preset information database.

[0057] In one embodiment, before obtaining the request information sent by the client, the server may generate guidance information, and the generated guidance information is used to guide the corresponding user of the client to send request information related to the topic indicated by the guidance information, wherein the correlation between the topic indicated by the guidance information and the target topic is greater than or equal to a preset correlation threshold. Specifically, Figure 4As shown, after the server establishes a communication connection with the client, the server may first generate guidance information based on the target topic, so that after sending the guidance information to the client, the client can be guided to send request information related to the topic indicated by the guidance information to the corresponding user, wherein the server may determine the target topic based on the keywords stored in each node in the knowledge graph and generate guidance information. For example, if the keywords stored in each node in the knowledge graph are words related to the World Cup, the determined target topic is the World Cup topic, and the generated guidance information is information for guiding the World Cup topic; or, the server may determine the target topic based on slot guidance and generate guidance information, wherein the slot refers to the key information that the server expects to collect from the user, and the slot guidance is guidance information determined based on the key information expected to be collected, such as the slot indicates that the key information expected to be collected is the new album directory of singer A, then the determined target topic is the album of singer A, and further, the generated guidance information may be information about the album of singer A; or, the server may also determine the target topic based on natural language understanding / natural language processing (i.e., NLP / NLU) technology and further generate guidance information, or, the server may also determine the target topic based on an artificial template or a chat module and generate guidance information.

[0058] After the server generates the guide information, the guide information may be sent to the client, and request information may be obtained from the client. After the server obtains the request information, the server may detect the subject indicated by the request information to determine whether the subject indicated by the request information is related to the target subject. Specifically, the server may detect whether the correlation between the subject indicated by the request information and the target subject is less than a preset correlation threshold. If it is determined to be less than, it is determined that the subject indicated by the request information is not related to the target subject. If it is determined to be greater than or equal to the correlation threshold, it is determined that the subject indicated by the request information is related to the target subject. In one embodiment, when the server determines that the topic indicated by the request information is not related to the target topic, it continues to generate guidance information and send it to the client until it receives request information related to the target topic. After the server receives the request information related to the target topic, the server can determine the response information corresponding to the request information from a preset information database to answer the request information. During multiple rounds of questions and answers between the server and the client, the server can collect the request information, or collect information pairs consisting of request information and response information, so as to enrich the server information data. In addition, the server can also update the original data in the server information database based on the collected information data. For the specific update process, please refer to the description of step S203 in the above embodiment.

[0059] In one embodiment, when the server receives the request information and determines that the relevance between the subject indicated by the request information and the target subject is less than a preset relevance threshold, the flow chart of the server generating the guide information can be seen in Figure 5 As shown in the figure, after the server obtains the request information from the client, it can first extract the request keywords from the request information. After the server extracts the request keywords, it can first use the short text named entity recognition technology to determine the entity indicated by the request keyword, so that the weight value of the entity indicated by each request keyword can be calculated, so as to determine the entity with the largest weight value. Further, after performing operations such as disambiguation on the entity with the largest weight value, the target entity corresponding to the maximum weight value can be determined. Among them, the entity includes keywords and the specific substance indicated by the keyword. The disambiguation operation refers to the process of determining the only substance indicated by the entity when the same entity indicates multiple different substances. For example, when the entity is an apple, the apple can indicate a substance with the attribute of fruit and a substance with the attribute of smart terminal respectively. Therefore, after determining that the entity includes apple based on the request keyword, the disambiguation operation can be used to determine the only substance indicated by apple in the request information (a substance with the attribute of fruit or a substance with the attribute of smart terminal). The short text named entity recognition technology may be, for example, one or more of a dictionary, statistics, a hidden Markov model, a language model, a maximum entropy model, a support vector machine (SVM), a conditional random field, a decision tree, and the English dictionary WordNet.

[0060] After the service determines the target entity, it can determine the subject indicated by the request information based on the target entity, or the server can directly process the request information using information subject identification technology to determine the subject indicated by the request information. After the server determines the subject indicated by the request information, it can determine whether the subject indicated by the request information is consistent with the target subject. If they are consistent, the server can search for the response keyword corresponding to the request keyword from the knowledge graph based on the request keyword, wherein the knowledge graph stores at least two keywords and the association relationship between any two keywords, so that the response information corresponding to the request information can be generated based on the response keyword and the response template in the information database.

[0061] In one embodiment, if it is determined that the subject indicated by the request information is inconsistent with the target subject, a search can be performed based on the full-domain knowledge graph to determine the response information corresponding to the request information from the Internet. After the server searches based on the full-domain knowledge graph, the response keywords corresponding to the request information can be determined based on the information pairs in the information database and the chat model. Furthermore, the response keywords can be filled in the response template. In addition, the server can also generate guidance information based on the chat module and slot guidance, so that the response keywords can be filled in the response template and then combined with the guidance information to form a response information output, thereby completing the answer to the request information and guiding subsequent request information.

[0062] In one embodiment, if the server fails to find the response keyword corresponding to the request keyword from the knowledge graph based on the request keyword, it may first search for similar information whose matching degree with the request information meets a preset matching degree threshold from the information database (i.e., information pair) based on the request keyword; further, the server may search for the response information corresponding to the similar information based on the information pair, and use the response information corresponding to the similar information as the response information corresponding to the request information. After determining the response information corresponding to the request information, the server may continue to obtain the request information from the client to implement information interaction with the client.

[0063] S302: Send the response information to the client, and obtain feedback information input based on the response information from the client.

[0064] In one embodiment, if the server successfully obtains the feedback information from the client, it goes on to execute step S303; if the server does not obtain the feedback information from the client, it can further determine whether the response information has an information tag, and the existing information tag is determined based on the historical feedback information input by each user in the user set for the response information; if the server determines that the response information has the information tag, it obtains the existing information tag, so that the response information can be optimized based on the existing information tag to obtain the optimized response information.

[0065] In one embodiment, if the server determines that the response information does not have the information tag, that is, the server has not obtained feedback information for the response information in history, the request information is clustered to determine the information set to which the request information belongs. The server may use a clustering algorithm to perform a clustering operation on the request information to determine the information set to which the request information belongs. Further, the server may use the target information tag of the target response information in the information set as the information tag of the response information. The information set includes at least one request information and the response information corresponding to the request information. The information categories of the information in the information set are the same. The target response information is any response information in the information set that has an information tag.

[0066] S303, determining the satisfaction with the response information based on the feedback information, and optimizing the response information when the satisfaction is less than a preset satisfaction threshold to obtain optimized response information, wherein the satisfaction determined based on the optimized response information is greater than or equal to the satisfaction threshold.

[0067] In one embodiment, before optimizing the response information, the server may first add an information tag to the response information, the information tag is used to indicate that the response information is optimized, and the information tag may be the above-mentioned expert tag. Specifically, the server may first determine the number of first users in the user set whose satisfaction with the response information is less than the satisfaction threshold, the user set includes a first user subset and a second user subset, the satisfaction of each user in the first user subset with the response information is less than the satisfaction threshold, and the satisfaction of each user in the second user subset with the response information is greater than or equal to the satisfaction threshold; further, the server may determine the proportion of users whose satisfaction with the response information is less than the satisfaction threshold based on the first number of users and the second number of users included in the second user subset; thereby, the server may add an information tag to the response information when the proportion of users is greater than or equal to a preset proportion threshold.

[0068] In one embodiment, after receiving feedback information from any user's corresponding client based on the response information, the server may first add a user tag to the response information based on the feedback information sent by the user's corresponding client. The user tag is used to indicate the user's satisfaction with the response information. Further, the server may determine the satisfaction of the user set with the response information based on a preset user set, thereby adding an information tag (i.e., an expert tag) to the response information based on the user set's satisfaction with the response information. After the server adds the expert tag to the response information, the response information with the expert tag added may be output to relevant optimization personnel, thereby optimizing the response information. After the server optimizes the response information to obtain the optimized response information, the server may turn to execute step S304.

[0069] S304: Use a preset natural language processing algorithm to perform word segmentation processing on the optimized response information, and extract at least two response keywords from the optimized response information.

[0070] S305: Determine the entity indicated by any one of the at least two response keywords, and determine the association relationship between any two of the at least two response keywords.

[0071] S306: Update the knowledge graph based on the entity indicated by any of the response keywords and the association relationship between any two of the response keywords.

[0072] In step S304 to step S306, the server may update the knowledge graph based on the optimized response information. Specifically, the server may use the NLP algorithm to perform word segmentation on the optimized response information, thereby extracting at least two response keywords. Further, the above-mentioned short text named entity recognition technology may be used to determine the entities indicated by the at least two response keywords, and the association relationship between any two response keywords, so that the knowledge graph may be updated based on the entity indicated by any keyword and the association relationship between any two keywords. Specifically, the server may first determine the storage location of any response keyword in the knowledge graph for any response keyword. Specifically, the server may first determine the storage node for storing any response keyword in the knowledge graph, so that the storage location of any response keyword may be determined according to the location of the storage node in the knowledge graph. In one embodiment, if the server does not determine the storage node for storing any response keyword from the knowledge graph, a new storage node for storing any response keyword is created in the knowledge graph, and the location of the newly created storage node in the knowledge graph is the storage location of any response keyword.

[0073] Furthermore, after determining the storage location of any response keyword in the knowledge graph, the server may store the any response keyword in the storage location. In addition, the server may also determine whether the association relationship between any two response keywords has changed, and when it is determined that the change has occurred, update the association relationship between any two response keywords. The update of the association relationship between any two keywords includes: adding a connecting edge for the any two associated keywords, or deleting the connecting edge between the two keywords.

[0074] In one embodiment, if the server uses an NLP algorithm to determine that the target keyword among the at least two keywords does not indicate an entity, the response information corresponding to the request information stored in the preset database is updated to the optimized response information, wherein the target keyword is any keyword among the at least two response keywords. For example, if the target keyword is "fatherly love", then the target keyword does not indicate an entity, the knowledge graph is not updated, and the response information corresponding to the request information stored in the database is updated to the optimized response information.

[0075] In an embodiment of the present invention, after obtaining the request information sent by the client, the server may send response information corresponding to the request information determined from a preset information database to the client, thereby obtaining feedback information input based on the response information from the client. The server may optimize the original response information based on the feedback information to obtain optimized response information. Further, the server may perform word segmentation processing on the optimized response information to obtain at least two response keywords, thereby updating the knowledge graph based on the entities indicated by the response keywords and the association between any two response keywords, thereby updating the original information data in the database and improving the update efficiency of the server. Based on the updated information database of the server, response information that makes the user more satisfied can be output, thereby improving the user's experience of using the server's question and answer system.

[0076] In one embodiment, Figure 6As shown, the request information obtained by the server from the client may be, for example, "Do you know the theme song of this World Cup?" Based on the search of the knowledge graph storing keywords related to the World Cup, the server may determine the response information corresponding to the request information and output it to the client. The response information output to the client may be, for example, "The theme song of the 2018 Russian World Cup is "Live It Up" sung by Diplo." Based on the response information, the feedback information obtained by the server is "Yes, I like the theme song of this World Cup very much." Based on the feedback information, the server may determine that the user's satisfaction with the response information is greater than the preset satisfaction threshold, and then the response information is not optimized. In addition, the server also obtains the request information "By the way, has Jay Chou released a new album?" Since the server determines based on the request information that the theme of the request information is not related to the target theme indicated by the knowledge graph, after determining the answer corresponding to the request information, it will also combine with the guidance information to generate response information, and send the response information to the client, such as the generated response information "Yes, the new album released in 2016 is called Bedtime Stories, Jay Chou's songs are very nice, but the World Cup is very popular recently, do you have any questions about the World Cup?" After determining the answer corresponding to the request information, the server may send the answer and the guidance information to the client separately or in combination, thereby realizing information interaction between the client and the server.

[0077] Based on the description of the above information optimization method embodiment, the present invention also proposes an information optimization device, which can be a computer program (including program code) running on the above server. The information optimization device can be used to perform the following steps: Figure 2 and Figure 3 For information optimization methods, see Figure 7 The information optimization device may include: an acquisition unit 701, a determination unit 702, a sending unit 703 and an optimization unit 704.

[0078] The acquisition unit 701 is used to acquire the request information sent by the client;

[0079] A determination unit 702 is used to determine the response information corresponding to the request information from a preset information database;

[0080] The sending unit 703 is used to send the response information to the client;

[0081] The acquisition unit 701 is further configured to acquire feedback information based on the response information input from the client;

[0082] The optimization unit 704 is used to determine the satisfaction level of the response information based on the feedback information, and when the satisfaction level is less than a preset satisfaction level threshold, optimize the response information to obtain optimized response information, wherein the satisfaction level determined based on the optimized response information is greater than or equal to the satisfaction level threshold.

[0083] In one embodiment, the device further includes: an adding unit 705 .

[0084] The adding unit 705 is used to add an information tag to the response information, where the information tag is used to indicate that the response information is optimized.

[0085] In one embodiment, the adding unit 705 is specifically used to:

[0086] Determine the number of first users in a user set whose satisfaction with the response information is less than the satisfaction threshold, the user set comprising a first user subset and a second user subset, the satisfaction of each user in the first user subset with the response information is less than the satisfaction threshold, and the satisfaction of each user in the second user subset with the response information is greater than or equal to the satisfaction threshold;

[0087] Determine, based on the first number of users and the second number of users included in the second user subset, a proportion of users whose satisfaction with the response information is less than the satisfaction threshold;

[0088] When the user proportion is greater than or equal to a preset proportion threshold, an information mark is added to the response information.

[0089] In one embodiment, the device further includes: a determining unit 706 .

[0090] A judging unit 706 is configured to judge whether the response information has an information tag if the obtaining unit 701 does not obtain the feedback information from the client, where the existing information tag is determined based on historical feedback information input by each user in the user set for the response information;

[0091] The acquiring unit 701 is further configured to acquire the existing information tag if it is determined that the response information has the information tag;

[0092] The optimization unit 704 is further configured to optimize the response information based on the existing information tag to obtain optimized response information.

[0093] In one embodiment, the determining unit 702 is further configured to perform a clustering operation on the request information to determine the information set to which the request information belongs if it is determined that the response information does not contain the information tag;

[0094] The determination unit 702 is further configured to use the target information tag of the target response information in the information set as the information tag of the response information, the information set includes at least one request information and response information corresponding to the request information, and each information in the information set belongs to the same information category.

[0095] In one embodiment, the device further includes: a word segmentation unit 707 and an updating unit 708 .

[0096] A word segmentation unit 707 is used to perform word segmentation processing on the optimized response information by using a preset natural language processing algorithm to extract at least two response keywords from the optimized response information;

[0097] The determining unit 702 is further configured to determine an entity indicated by any one of the at least two response keywords, and determine an association relationship between any two of the at least two response keywords;

[0098] The updating unit 708 is used to update the knowledge graph based on the entity indicated by any response keyword and the association relationship between any two response keywords.

[0099] In one embodiment, the updating unit 708 is specifically configured to:

[0100] For any response keyword, determine the storage location of the any response keyword in the knowledge graph, and store the any response keyword in the storage location;

[0101] It is determined whether the association relationship between the two arbitrary response keywords has changed, and when it is determined that the association relationship between the two arbitrary response keywords has changed, the association relationship between the two arbitrary response keywords is updated.

[0102] In one embodiment, the determining unit 702 is further configured to determine that the target keyword in the at least two response keywords does not have an entity indicated if a preset natural language processing algorithm is used;

[0103] The updating unit 708 is further configured to update the response information corresponding to the request information stored in the information database to the optimized response information.

[0104] In one embodiment, the determining unit 702 is specifically configured to:

[0105] Obtaining request information sent by the client, and extracting request keywords from the request information;

[0106] Based on the request keyword, searching for a response keyword corresponding to the request keyword from a knowledge graph, wherein the knowledge graph stores at least two response keywords and an association relationship between any two response keywords;

[0107] Based on the response keyword and the response template in the information database, response information corresponding to the request information is generated.

[0108] In one embodiment, the device further includes: a searching unit 709 .

[0109] A search unit 709 is configured to search the information database for similar information whose matching degree with the request information meets a preset matching degree threshold based on the request keyword if a response keyword corresponding to the request keyword is not found in the knowledge graph based on the request keyword;

[0110] The searching unit 709 is further configured to search for response information corresponding to the similar information, and use the response information corresponding to the similar information as the response information corresponding to the request information.

[0111] In one embodiment, the device further includes: a generating unit 710 .

[0112] A generating unit 710 is configured to generate guidance information, wherein the guidance information is generated before obtaining the request information sent by the client, or the guidance information is generated when the degree of association between the subject indicated by the request information and the target subject is less than a preset association threshold, and the degree of association between the subject indicated by the guidance information and the target subject is greater than or equal to the association threshold;

[0113] The sending unit 703 is further configured to send the guide information to the client, so as to guide the client to send request information whose relevance to the subject indicated by the guide information is greater than or equal to the relevance threshold.

[0114] In an embodiment of the present invention, the acquisition unit 701 can acquire the request information sent by the client, so that the determination unit 702 can determine the response information corresponding to the request information from a preset information database. After the sending unit 703 sends the determined response information to the client, the acquisition unit 701 can acquire feedback information input based on the response information from the client, so that after the determination unit 702 determines the satisfaction with the response information according to the feedback information, when it is determined that the satisfaction is less than a preset satisfaction threshold, the optimization unit 704 can optimize the response information, thereby obtaining response information that can improve user satisfaction, thereby realizing automatic updating of the response information in the server according to user feedback, and improving the updating efficiency of the response information.

[0115] See also Figure 8 , is a schematic block diagram of the structure of a server provided by an embodiment of the present invention. Figure 8 The server in the embodiment shown may include: one or more processors 801; one or more input devices 802, one or more output devices 803 and a memory 804. The processor 801, the input device 802, the output device 803 and the memory 804 are connected via a bus 805. The memory 804 is used to store a computer program, the computer program includes program instructions, and the processor 801 is used to execute the program instructions stored in the memory 804.

[0116] The memory 804 may include a volatile memory (volatile memory), such as a random-access memory (RAM); the memory 804 may also include a non-volatile memory (non-volatile memory), such as a flash memory, a solid-state drive (SSD), etc.; the memory 804 may also include a combination of the above types of memory.

[0117] The processor 801 may be a central processing unit (CPU). The processor 801 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), etc. The PLD may be a field-programmable gate array (FPGA), a generic array logic (GAL), etc. The processor 801 may also be a combination of the above structures.

[0118] In the embodiment of the present invention, the memory 804 is used to store a computer program, the computer program includes program instructions, and the processor 801 is used to execute the program instructions stored in the memory 804 to implement the above-mentioned Figure 2 and Figure 3 The steps of the corresponding method in .

[0119] In one embodiment, the processor 801 is configured to call the program instructions to execute:

[0120] Obtaining the request information sent by the client, and determining the response information corresponding to the request information from a preset information database;

[0121] Sending the response information to the client, and obtaining feedback information input based on the response information from the client;

[0122] The satisfaction level for the response information is determined based on the feedback information, and when the satisfaction level is less than a preset satisfaction level threshold, the response information is optimized to obtain optimized response information, wherein the satisfaction level determined based on the optimized response information is greater than or equal to the satisfaction level threshold.

[0123] In one embodiment, the processor 801 is further configured to call the program instructions to execute:

[0124] An information tag is added to the response information, where the information tag is used to indicate that the response information is optimized.

[0125] In one embodiment, the processor 801 is further configured to call the program instructions to execute:

[0126] Determine the number of first users in a user set whose satisfaction with the response information is less than the satisfaction threshold, the user set comprising a first user subset and a second user subset, the satisfaction of each user in the first user subset with the response information is less than the satisfaction threshold, and the satisfaction of each user in the second user subset with the response information is greater than or equal to the satisfaction threshold;

[0127] Determine, based on the first number of users and the second number of users included in the second user subset, a proportion of users whose satisfaction with the response information is less than the satisfaction threshold;

[0128] When the user proportion is greater than or equal to a preset proportion threshold, an information mark is added to the response information.

[0129] In one embodiment, the processor 801 is further configured to call the program instructions to execute:

[0130] If the feedback information is not obtained from the client, determining whether the response information has an information tag, where the existing information tag is determined based on historical feedback information input by each user in the user set for the response information;

[0131] If it is determined that the response information has the information tag, the existing information tag is obtained, and the response information is optimized based on the existing information tag to obtain the optimized response information.

[0132] In one embodiment, the processor 801 is further configured to call the program instructions to execute:

[0133] If it is determined that the response information does not contain the information tag, clustering the request information to determine the information set to which the request information belongs;

[0134] The target information tag of the target response information in the information set is used as the information tag of the response information. The information set includes at least one request information and response information corresponding to the request information. The information categories of each information in the information set are the same.

[0135] In one embodiment, the processor 801 is further configured to call the program instructions to execute:

[0136] Using a preset natural language processing algorithm to perform word segmentation processing on the optimized response information, and extracting at least two response keywords from the optimized response information;

[0137] Determine an entity indicated by any one of the at least two response keywords, and determine an association relationship between any two of the at least two response keywords;

[0138] The knowledge graph is updated based on the entity indicated by any of the response keywords and the association relationship between any two of the response keywords.

[0139] In one embodiment, the processor 801 is further configured to call the program instructions to execute:

[0140] For any response keyword, determine the storage location of the any response keyword in the knowledge graph, and store the any response keyword in the storage location;

[0141] It is determined whether the association relationship between the two arbitrary response keywords has changed, and when it is determined that the association relationship between the two arbitrary response keywords has changed, the association relationship between the two arbitrary response keywords is updated.

[0142] In one embodiment, the processor 801 is further configured to call the program instructions to execute:

[0143] If a preset natural language processing algorithm is used, it is determined that the target keyword in the at least two response keywords does not have an entity indicated;

[0144] The response information corresponding to the request information stored in the information database is updated to the optimized response information.

[0145] In one embodiment, the processor 801 is further configured to call the program instructions to execute:

[0146] Obtaining request information sent by the client, and extracting request keywords from the request information;

[0147] Based on the request keyword, searching for a response keyword corresponding to the request keyword from a knowledge graph, wherein the knowledge graph stores at least two response keywords and an association relationship between any two response keywords;

[0148] Based on the response keyword and the response template in the information database, response information corresponding to the request information is generated.

[0149] In one embodiment, the processor 801 is further configured to call the program instructions to execute:

[0150] If a response keyword corresponding to the request keyword is not found in the knowledge graph based on the request keyword, similar information whose matching degree with the request information meets a preset matching degree threshold is searched from the information database based on the request keyword;

[0151] Response information corresponding to the similar information is searched, and the response information corresponding to the similar information is used as the response information corresponding to the request information.

[0152] In one embodiment, the processor 801 is further configured to call the program instructions to execute:

[0153] Generate guidance information, where the guidance information is generated before obtaining the request information sent by the client, or the guidance information is generated when the correlation between the subject indicated by the request information and the target subject is less than a preset correlation threshold, and the correlation between the subject indicated by the guidance information and the target subject is greater than or equal to the correlation threshold;

[0154] The guide information is sent to the client to guide the client to send request information whose relevance to the subject indicated by the guide information is greater than or equal to the relevance threshold.

[0155] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0156] What is disclosed above is only a partial embodiment of the present invention, which certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of implementing the above embodiments and making equivalent changes according to the claims of the present invention still fall within the scope of the invention.

Claims

1. An information optimization method, characterized in that: include: Establishing a communication connection with a client through a preset communication interface triggers the generation of a system account for accessing the client, and the system account is used to exchange information with a user account of the client; a preset information database is encapsulated in the communication interface, and the information database includes a knowledge graph, an information pair, and a chat module; Obtaining request information sent by the client based on the system account, and extracting request keywords from the request information; Based on the request keyword, searching the knowledge graph for a response keyword corresponding to the request keyword, wherein the knowledge graph stores at least two response keywords and an association relationship between any two response keywords; generating response information corresponding to the request information based on the response keyword and the response template in the information database; If a response keyword corresponding to the request keyword is not found from the knowledge graph based on the request keyword, similar information whose matching degree with the request information meets a preset matching degree threshold is searched from the information pair based on the request keyword; Searching for response information corresponding to the similar information, and using the response information corresponding to the similar information as the response information corresponding to the request information; Sending the response information to the client, and obtaining feedback information input based on the response information from the client; determining satisfaction with the response information according to the feedback information; Determine the number of first users in a user set whose satisfaction with the response information is less than a preset satisfaction threshold, the user set comprising a first user subset and a second user subset, the satisfaction of each user in the first user subset with the response information is less than the satisfaction threshold, and the satisfaction of each user in the second user subset with the response information is greater than or equal to the satisfaction threshold; Determine, based on the first number of users and the second number of users included in the second user subset, a proportion of users whose satisfaction with the response information is less than the satisfaction threshold; When the user proportion is greater than or equal to a preset proportion threshold, adding an information mark to the response information, wherein the information mark is used to indicate that the response information is optimized; According to the indication of the information tag, the response information is optimized to obtain optimized response information, wherein the satisfaction level determined according to the optimized response information is greater than or equal to the satisfaction level threshold; If the similar information is not found from the information pair based on the request keyword, detecting the subject indicated by the request information, and determining the degree of association between the subject indicated by the request information and the target subject; If the relevance is less than a preset relevance threshold, generating guidance information through the small talk module, the relevance between the topic indicated by the guidance information and the target topic is greater than or equal to the relevance threshold; The guide information is sent to the client, and request information sent based on the guide information and having a relevance greater than or equal to the relevance threshold value is obtained from the client.

2. The method according to claim 1, characterized in that The method further comprises: If the feedback information is not obtained from the client, determining whether the response information has an information tag, where the existing information tag is determined based on historical feedback information input by each user in the user set for the response information; If it is determined that the response information has the information tag, the existing information tag is obtained, and the response information is optimized based on the existing information tag to obtain the optimized response information.

3. The method according to claim 2, characterized in that The method further comprises: If it is determined that the response information does not contain the information tag, clustering the request information to determine the information set to which the request information belongs; The target information tag of the target response information in the information set is used as the information tag of the response information. The information set includes at least one request information and response information corresponding to the request information. The information categories of each information in the information set are the same.

4. The method according to claim 1, characterized in that: After obtaining the optimized response information, the method further includes: Using a preset natural language processing algorithm to perform word segmentation processing on the optimized response information, and extracting at least two response keywords from the optimized response information; Determine an entity indicated by any one of the at least two response keywords, and determine an association relationship between any two of the at least two response keywords; The knowledge graph is updated based on the entity indicated by any of the response keywords and the association relationship between any two of the response keywords.

5. The method according to claim 4, characterized in that The updating of the knowledge graph based on the entity indicated by any one of the response keywords and the association relationship between any two of the response keywords includes: For any response keyword, determine the storage location of the any response keyword in the knowledge graph, and store the any response keyword in the storage location; It is determined whether the association relationship between the two arbitrary response keywords has changed, and when it is determined that the association relationship between the two arbitrary response keywords has changed, the association relationship between the two arbitrary response keywords is updated.

6. The method according to claim 4, characterized in that The method further comprises: If a preset natural language processing algorithm is used, it is determined that the target keyword in the at least two response keywords does not have an entity indicated; The response information corresponding to the request information stored in the information database is updated to the optimized response information.

7. The method according to claim 1, characterized in that The method further comprises: Generate guidance information, where the guidance information is generated before obtaining the request information sent by the client; The guide information is sent to the client to guide the client to send request information whose relevance to the subject indicated by the guide information is greater than or equal to the relevance threshold.

8. An information optimization device, characterized in that: include: An acquisition unit, configured to establish a communication connection with a client through a preset communication interface, trigger the generation of a system account for accessing the client, and acquire request information sent by the client based on the system account, wherein a preset information database is encapsulated in the communication interface, and the information database includes a knowledge graph, an information pair, and a chat module, and the system account is used to exchange information with a user account of the client; A determination unit, configured to determine response information corresponding to the request information from a preset information database; The determining unit is specifically configured to extract a request keyword from the request information, and search the knowledge graph for a response keyword corresponding to the request keyword based on the request keyword, wherein the knowledge graph stores at least two response keywords and an association relationship between any two response keywords; generating response information corresponding to the request information based on the response keyword and the response template in the information database; A search unit, configured to search from the information pair for similar information whose matching degree with the request information satisfies a preset matching degree threshold based on the request keyword, if a response keyword corresponding to the request keyword is not found from the knowledge graph based on the request keyword; Searching for response information corresponding to the similar information, and using the response information corresponding to the similar information as the response information corresponding to the request information; A sending unit, configured to send the response information to the client; The acquisition unit is further configured to acquire feedback information input based on the response information from the client; an adding unit, configured to determine satisfaction with the response information according to the feedback information, and determine the number of first users in a user set whose satisfaction with the response information is less than a preset satisfaction threshold, the user set comprising a first user subset and a second user subset, the satisfaction of each user in the first user subset with the response information is less than the satisfaction threshold, and the satisfaction of each user in the second user subset with the response information is greater than or equal to the satisfaction threshold; Determine, based on the first number of users and the second number of users included in the second user subset, a proportion of users whose satisfaction with the response information is less than the satisfaction threshold; When the user proportion is greater than or equal to a preset proportion threshold, adding an information mark to the response information, wherein the information mark is used to indicate that the response information is optimized; an optimization unit, configured to optimize the response information according to an indication of the information tag to obtain optimized response information, wherein a satisfaction level determined according to the optimized response information is greater than or equal to the satisfaction level threshold; a generating unit, configured to detect the subject indicated by the request information and determine the degree of association between the subject indicated by the request information and a target subject if the similar information is not found from the information pair based on the request keyword; if the degree of association is less than a preset degree of association threshold, generate guiding information through the small talk module, wherein the degree of association between the subject indicated by the guiding information and the target subject is greater than or equal to the degree of association threshold; The sending unit is further configured to send the guide information to the client to guide the client to send request information whose relevance to the subject indicated by the guide information is greater than or equal to the relevance threshold.

9. A server, characterized in that: The method comprises a processor, a memory and a communication interface, wherein the processor, the memory and the communication interface are connected to each other, wherein the memory is used to store computer program instructions, and the processor is configured to execute the program instructions to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor is caused to perform the method according to any one of claims 1 to 7.

Images